Multistage production systems with random yield: Heuristics and optimality

Most previous work on adjusting lot-sizes to recognize yield losses has been restricted to single-stage production processes. Here the paper examines multistage serial or linear production systems with yield loss. After developing a dynamic programming optimization to be used as a benchmark, four single-stage, lot-sizing heuristics are adapted for use in the multistage system, and their performance over a 320-problem test bed is compared with the optimization. Two heuristics based on marginal analysis are found to give close-to-optimal results.